Development of immortalized Hertwig’s epithelial root ...cementum-forming cells via the process of epithelial– mesenchymal transition (EMT) [7, 8]. With the elong-ation of HERS,

RESEARCH Open Access

Development of immortalized Hertwig’sepithelial root sheath cell lines forcementum and dentin regenerationXuebing Li1,2,3,4†, Sicheng Zhang1,2,3†, Zirui Zhang1,2,3,6, Weihua Guo1,2,3,5, Guoqing Chen1,2,3* andWeidong Tian1,2,3,4*

Abstract

Background: Hertwig’s epithelial root sheath (HERS) is important in guiding tooth root formation by differentiatinginto cementoblasts through epithelial–mesenchymal transition (EMT) and inducing odontoblastic differentiation ofdental papilla through epithelial–mesenchymal interaction (EMI) during the tooth root development. Thus, HERScells are critical for cementum and dentin formation and might be a potential cell source to achieve tooth rootregeneration. However, limited availability and lifespan of primary HERS cells may represent an obstacle for biologicalinvestigation and therapeutic use of tooth tissue engineering. Therefore, we constructed, characterized, and tested thefunctionality of immortalized cell lines in order to produce a more readily available alternative to HERS cells.

Methods: Primary HERS cells were immortalized via infection with lentivirus vector containing the gene encodingsimian virus 40 Large T Antigen (SV40LT). Immortalized HERS cell subclones were isolated using a limiting dilutionmethod, and subclones named HERS-H1 and HERS-C2 cells were isolated. The characteristics of HERS-H1 and HERS-C2cells, including cell proliferation, ability of epithelial–mesenchymal transformation and epithelial–mesenchymalinteraction, were determined by CCK-8 assay, immunofluorescence staining, and real-time PCR. The cell differentiationinto cementoblast-like cells or periodontal fibroblast-like cells was confirmed in vivo. And the inductive influence of thecell lines on dental papilla cells (DPCs) was also confirmed in vivo.

Results: HERS-H1 and HERS-C2 cells share some common features with primary HERS cells such as epithelial-likemorphology, positive expression of CK14, E-Cadherin, and Vimentin, and undergoing EMT in response to TGF-beta. HERS-C2 cells showed the EMT characteristics and could differentiate into cementum-forming cells in vitroand generate cementum-like tissue in vivo. HERS-H1 could induce the differentiation of DPCs into odontoblastsin vitro and generation of dentin-like tissue in vivo.

Conclusions: We successfully isolated and characterized novel cell lines representing two key features of HERScells during the tooth root development and which were useful substitutes for primary HERS cells, therebyproviding a biologically relevant, unlimited cell source for studies on cell biology, developmental biology, andtooth root regeneration.

Keywords: Hertwig’s epithelial root sheath, Cementum, Dentin, Tooth regeneration

* Correspondence: [email protected]; [email protected]†Xuebing Li and Sicheng Zhang contributed equally to this work.1State Key Laboratory of Oral Disease, West China Hospital of Stomatology,Sichuan University, Chengdu, ChinaFull list of author information is available at the end of the article

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Li et al. Stem Cell Research & Therapy (2019) 10:3 https://doi.org/10.1186/s13287-018-1106-8

BackgroundTooth loss is a common and frequent situation resultedfrom numerous pathologies such as trauma, endodonticcomplications, and periodontal and carious diseases.Due to the limitations of traditional artificial repair oftooth loss, stem cell-based tissue engineering is consid-ered a valid therapeutic approach to construct a toothwith normal physiological functions [1]. There have beenvarious kinds of mesenchymal stem cells for generatingbioengineered tooth tissues, such as dental pulp cells,dental follicle cells (DFCs), and bone marrow stromalcells (BMSCs) [2–4]. From a developmental point ofview, the development and growth of a functional toothare complex processes, in which reciprocal interactionbetween mesenchymal and epithelial cells plays an im-portant role [1]. Although significant progress has beenmade with mesenchymal cells in tooth tissue engineer-ing, there is little information available for dental epithe-lial stem cells.The Hertwig’s epithelial root sheath (HERS) is a bi-

layer epithelial sheath originated from the apical regionof the enamel organ [5]. As for the process of tooth rootdevelopment, HERS plays an important role in guidingroot formation and determining the size, shape, andnumber of tooth roots [6]. During the tooth rootdevelopment, HERS cells can differentiate intocementum-forming cells via the process of epithelial–mesenchymal transition (EMT) [7, 8]. With the elong-ation of HERS, the dental papilla comes to contact withHERS and differentiate into odontoblasts, which secretedentin of the root. Moreover, the HERS also producesgrowth factors that contribute to the induction ofodontoblast differentiation [6]. If the structure of HERSis disturbed prematurely, the differentiation of rootodontoblasts is compromised [9]. Thus, it is confirmedthat HERS functions as a development center to guideroot formation. As for the mechanism involved in theHERS cells or odontoblast differentiation, our and otherprevious studies had shown that TGF-β1 or FGF2 couldtrigger the EMT of HERS cells, in which PI3K/AKT andMAPK/ERK signaling was involved [7, 8]. BMP/TGF-βsignaling pathway was mainly discussed in the differenti-ation of odontoblasts during tooth root development[10–13]. However, HERS is a fleeting, transient structureassembled in the early period of root formation andelongation, which makes it difficult to obtain enoughprimary cells for biological and preclinical studies. Withthe root formation, HERS cells fenestrated and reducedto epithelial rests of Malassez (ERM) after root dentindeposition. There have been studies speculated thatERM could develop into cementum-forming cells andreported that immortalized cell lines had already beenestablished [7, 14, 15]. Nevertheless, these cells are ter-minal products of HERS, which may not completely

reflect the original characteristics of HERS cells in devel-oping tooth. Therefore, most aspects of the biology ofHERS cells are still obscure because of the lack of effi-cient methods to isolate stable phenotype and enoughprimary cells.As mentioned above, the HERS cells play a critical role

during the tooth root development via contributing tocell differentiation. However, there have been few reportsconcerned with the use of HERS cells in stem cell-basedapproaches of tooth tissue engineering. In this context,HERS cells might be a potential source for the construc-tion of tooth root. In this study, we isolated primaryHERS cells from SD rat and successfully established twoimmortalized cell lines with the characteristics of under-going EMT and conducting epithelial–mesenchymalinteraction (EMI) respectively. Importantly, we certifiedtheir potential of differentiation into cementoblasts andthe generation of cementum-like tissue in vivo, and in-duction of differentiation of dental papilla cells (DPCs)and the formation of dentin-like tissue were also evalu-ated in vivo. Our data indicates that immortalized HERScells can provide substitutes for primary cells in bio-logically relevant applications in the fields of tooth rootregeneration.

Materials and methodsAll experiments were conducted in accordance with aprotocol approved by the Committee of Ethics of WestChina Hospital of Stomatology of Sichuan University(NO. WCHSIRB-D-2018-100).

Primary cell isolation and lentiviral transductionThe primary DPCs were obtained and cultured as previ-ously described [16]. For the primary HERS cell isolationand culture, as shown in Fig. 1a, Hertwig’s epithelial rootsheath of the tooth apical region of the mandibular firstmolars at PN8 was dissected, the tissues were dissociatedwith 2.4 U/mL dispase (Sigma-Aldrich, St Louis, MO)and 625 U/mL collagenase (Sigma-Aldrich, St Louis,MO), which was cultured with epithelial cell medium(ScienCell, CA, USA) consisting of basal medium, 2%fetal bovine serum, 1% epithelial cell growth supplement,and 1% penicillin/streptomycin solution [8, 17]. Purifiedprimary HERS cells at passage 1 were transfected withlentiviral vector (pGMLV-SV40T-PURO, Genomeditech,Shanghai, China), which encoded simian virus 40 LargeT Antigen (SV40 LT) and a puromycin resistance gene.After transfection, the cells were selected by supple-menting the media with 2 μg/ml puromycin (Sigma-Al-drich, St Louis, MO). Using the limited dilution method,cells were isolated as single cells and cultured to expandthe selected clones.

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 2 of 13

Cell proliferation assayCell proliferation was detected using Cell CountingKit-8 (Dojindo, Japan) according to the manufacturer’sinstructions. Cells at passage 2 were plated at a densityof 5000 cells per well in 96-well plates with the completemedia. The original cultivation medium was replaced by120 μl fresh medium containing 12 μl CCK-8 at the sametime of consecutive 7 days. After 2 h incubation at 37 °C,the absorbance of the solution was determined with aspectrophotometer (Thermo Scientific, USA) at 450 nm.Five independent experiments were performed in eachgroup.

Real-time PCRCells at passage 4 were in an undifferentiated status andtreated with 10 ng/ml TGF-β1. RNA was isolated fromcells using RNAiso Plus (Takara, Japan) according to the

manufacturer’s protocol. Four micrograms of RNA werereverse transcribed to cDNA with Revert Aid FirstStrand cDNA Synthesis Kit Transcription Kit (ThermoScientific, USA). Real-time RT-PCR was conducted aspreviously described using SYBR Premix Ex Taq(TaKaRa Biotechnology, Japan) [16]. The reaction wasperformed with QuantStudio 6 Flex (Applied Biosys-tems, Foster City, CA, USA). All of the operating proce-dures followed the manufacturer’s protocol. The PCRprimer sequences are shown in Additional file 1: TableS1. All experiments were performed in triplicates and re-peated three times.

Immunofluorescence and immunohistochemical stainingImmunofluorescence and immunohistochemical stainingwas performed as previously described [8, 18]. Cells ortissue sections were examined under fluorescence

Fig. 1 Establishment of the immortalized HERS cell lines and they expressed markers specific to HERS cells. a Isolation of HERS cells from PN8 ratfirst mandibular molars. b The primary HERS cells showed a typical cobblestone-like appearance. c SV40 T-Ag was present in the nucleus of theimmortalized HERS cells. d Compared with the primary cells, the two cell lines showed an epithelial morphology and are a little bit elongated.e CCK-8 assay was used to compare the proliferation rate of the primary cells, HERS-H1 and HERS-C2. f The primary HERS cells expressed bothepithelial cell markers CK14, E-cadherin, and mesenchymal cell markers vimentin. The immortalized cells HERS-C2 and HERS-H1 also expressedCK14, E-cadherin, and vimentin stably at least from passage 2 to passage 20. Scale bars are shown. (*P < 0.05, HERS-H1 vs. primary HERS cells;#P < 0.05, HERS-C2 vs. primary HERS cells, Scale bar: 100 μm in b and d)

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 3 of 13

confocal microscope (Olympus FV1000, Japan) ormicroscope (Olympus, IX73, Japan). The primary andsecondary antibodies and their dilutions used in thisstudy were as follows: mouse anti-CK14 (1:200,MAB3232, Millipore, CA, USA), mouse anti-vimentin(1:200, sc-6260, Santa Cruz, CA, USA), ratanti-E-cadherin (1:100, ab11512, Abcam, MA, USA),rabbit anti-SV40 Large T antigen (1:200, #15729, CellSignaling Technology, CA, USA), goat anti-DMP1(1:200, sc-6551, Santa Cruz, CA, USA), rabbit anti-BSP(1:200, D261488, Sangon Biotech, Shanghai, China),rabbit anti-OCN (1:200, AP2002a 614487, Zenbio,Chengdu, China), rabbit anti-Periostin (1:200, sc-67233,Santa Cruz, CA, USA), rabbit anti-DSP (1:200, sc-33587,Santa Cruz, CA, USA), Alexa FluoR 488 Goat antiMouse (1:500, A11001, Invitrogen, Eugene, OR, USA),Alexa FluoR 488 Goat anti Rabbit (1:500, A11008, Invi-trogen, Eugene, OR, USA), Alexa FluoR 555 Goat antiRabbit (1:500, A11008, Invitrogen, Eugene, OR, USA),donkey anti-goat IgG-HRP (1:500, sc-2020, Santa Cruz,CA, USA), and goat anti-rabbit IgG-HRP (1:500,ZB-2301, ZSGB-BIO, Beijing, China). Negative controlswere carried out by substituting a normal IgG for theprimary antibody.

Mineralization assayCells were cultured in osteo-inductive medium with orwithout 10 ng/mL TGF-β1. After 3 and 7 days culture,Total RNA was isolated from cells for quantitativereal-time PCR analysis. After 4 weeks culture, the cellswere fixed with 4% paraformaldehyde and stained with0.1% Alizarin red S (Sigma-Aldrich, St Louis, MO) for 30min at room temperature. Mineralized bone nodules weredestained with 10% cetylpyridinium chloride in doubledistilled water, and the calcium concentration was deter-mined by absorbance measurements at 405 nm. All exper-iments were performed at least in triplicate.

Conditioned medium preparation and treatmentThe conditioned medium (CM) of the two cell lines wasprepared as described previously [19]. The DPCs werecultured with HERS-H1 CM or HERS-C2 CM andrefreshed every 24 h. At a different time point, the cellsco-cultured with CM were collected to performreal-time PCR. All experiments were performed at leastin triplicate.

In vivo transplantation of cultured cellsHERS-C2 and HERS-H1 cells (treated with 10 ng/mLTGF-β1 for 48 h) alone or combined with DPCs(HERS-C2 or HERS-H1: DPCs = 1:2, a total of approxi-mately 2.0 × 106 cells) were mixed with 40 mg hydroxy-apatite/tricalcium phosphate (HA/TCP) ceramic powder(The Engineering Research Center in Biomaterials,

Sichuan University), and two pellets of HA/TCP withadherent cells was then transplanted into left renal cap-sule of 8-week female SD rats (Dashuo ExperimentalAnimal Co. Ltd., Chengdu, China). The 24 rats wererandomly divided into six groups: Control group (HA/TCP without cells), HERS-C2 group (HERS-C2 cellsmixed with HA/TCP), HERS-H1 group (HERS-H1 cellsmixed with HA/TCP), DPC group (DPCs mixed withHA/TCP), HERS-C2 combined with DPC group(HERS-C2 combined with DPCs mixed with HA/TCP),and HERS-H1 combined with DPC group (HERS-H1combined with DPCs mixed with HA/TCP). There wereat least four replicates in each group. All animals weremaintained under standardized conditions with thetemperature of 21 °C and a 12-h light cycle and had freeaccess to food and water. The grafts were obtained at 8weeks post-operatively, fixed with 4% paraformaldehyde,decalcified with buffered 10% EDTA, and then embed-ded in paraffin for further experiments.

Tumorigenicity assayTo determine whether the HERS-C2 and HERS-H1 cellswere tumorigenic, 1.0 × 105 cells combined with Matrigel(BD Biosciences, USA) were subcutaneously injectedinto female 4–5-week-old balb/c nude mice (18–20 g)(Vital River Laboratory Animal Technology Co., Ltd.,Beijing, China). The tumorigenic and immortalized oralsquamous carcinoma SCC-25 cells were used as positivecontrol. The mice were randomly assigned to five groups(n = 4 mice/group): control group (sham); primary HERScells group; HERS-C2 cells group, HERS-H1 cells group,and SCC25 cells group. Animals were maintained underspecific pathogen-free (SPF) facility with access ad libitumto food and fresh water and were monitored for tumorformation. Mice were sacrificed after 4 weeks culture andtransplanted parts were obtained for further study.

Microarrays and gene expression analysisAgilent gene expression arrays (Kangchen Bio-tech Inc.,Shanghai. China) were used to investigate the global ex-pression profile of primary HERS cells, HERS-H1 andHERS-C2 cells (n = 3). The array represented more than41,000 transcripts. The records of microarrays data havebeen approved and assigned the accession number GEO:GSE109622. The microarray datasets were normalized inGeneSpring GX v12.1 software package (Agilent Tech-nologies). A p value of less than 0.01 and a mean expres-sion change of greater than twofold was consideredstatistically significant and these genes were used for fur-ther analysis. Gene ontology and signaling pathway ana-lysis of significantly different genes were analyzed usingthe DAVID online analysis tool (http://david.abcc.n-cifcrf.gov/).

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 4 of 13

Statistical analysisAll data were expressed as mean value ± standard devi-ation for each group. Statistical significance was assessedby using Student’s t test for two groups or analysis ofvariance (Tukey’s test) for multiple groups. P < 0.05 wasconsidered as statistically significant.

ResultsPhenotypic characteristics of two immortalized HERS celllines HERS-C2 and HERS-H1The primary HERS cells showed a cobblestone appear-ance (Fig. 1b). These cells were transfected with lenti-viral vector encoding SV40 LT and selected withpuromycin. Immunofluorescence detection showed theimmortalized cell lines were positive expression of SV40T-Ag in the nuclear (Fig. 1c). By selection for clonogeniccells, a total of 68 clones were selected which could becultured for more than 50 passages. Among them, thetwo cell lines named HERS-H1 and HERS-C2 were usedin present study for their unique features. These twotype cells, showing a cobblestone-like morphology(Fig. 1d), were adherent and had a high proliferationcapacity, with a doubling time of about 24 h (Fig. 1e). Allthe cells were positive for epithelial markers cytokeratin14 (CK14) and E-cadherin and mesenchymal markervimentin, which suggested the cells maintained the char-acteristics of both epithelial and mesenchymal cells.HERS-C2 and HERS-H1 maintained the expression ofHERS cells markers at least 20 passages (Fig. 1f ).To determine if the immortalized HERS-C2 and

HERS-H1 cells were tumorigenic, they were injectedsubcutaneously into immunodeficient athymic mice. Notumor formation was observed after 4 weeks, whereasthe SCC-25 tumor cells formed large tumors withinmuch shorter time (Additional file 2: Figure S1a andS1b).

EMT characteristics of HERS-C2 and HERS-H1 cellsIn previous studies, primary HERS cells could undergoEMT and acquire a mesenchymal phenotype with the in-duction of TGF-β1 [7, 8]. To investigate the propertiesof EMT of the immortalized cell lines, HERS-C2 andHERS-H1 were treated with TGF-β1 for 3 days and 7days. TGF-β1 treatment triggered a partial morpho-logical alteration of HERS-C2 and HERS-H1, from typ-ical cobblestone-like epithelial cells to spindle-shapemesenchymal-like cells (Fig. 2a). After 3 days treatment,the expression of epithelial-associated gene E-cadherinwas decreased while the mesenchymal-associated genesincluding vimentin and N-cadherin were increased inHERS-C2 cells (Fig. 2b). The transcription factors twist1,snail1, and zeb1 were upregulated (Fig. 2b). As forHERS-H1 cells, the expression of epithelial-associated geneE-cadherin was upregulated at 3 days after TGF-β1

treatment (Fig. 2c) and downregulated until 7 days afterTGF-β1 treatment (Additional file 3: Figure S2); the expres-sion levels of vimentin and N-cadherin were increased after3 days or 7 days treatment (Fig. 2c and Additional file 3:Figure S2). The transcription factors twist1, snail1, andzeb1 were upregulated (Fig. 2c). These data suggested thatimmortalized HERS cells could respond to TGF-β1 andacquire mesenchymal phenotypes through EMT.

Potential of cementoblastic differentiation of HERS-C2and HERS-H1 cells in vitro and in vivoTo address the potential of the two cell lines to undergocementoblastic differentiation, we cultured the cells withosteo-medium and evaluated the expression of markersof cementoblast such as bone sialoprotein (BSP), dentinmatrix protein 1 (DMP1), and collagen type 1A1(COL1A1) [20–23]. Osteogenic induction significantlyupregulated the expression of BSP and DMP1 but notaffected the COL1A1 expression at 3 days in HERS-H1and HERS-C2 cells (Fig. 3a, b).The differentiation of HERS cells into cementoblasts

was also a process of EMT [7], so we added TGF-β1 intothe osteogenic medium to investigate whether TGF-β1could promote osteogenic differentiation. Compared toosteogenic induction alone, TGF-β1 treatment combinedwith osteogenic induction significantly upregulated theBSP and DMP1 expression at 7 days and COL1A1 ex-pression at 3 days in HERS-H1 cells (Fig. 3a). InHERS-C2 cells, TGF-β1 treatment combined with osteo-genic induction significantly upregulated the BSP andDMP1 expression from 3 days to 7 days, and COL1A1 ex-pression at 7 days (Fig. 3b). Alizarin red S staining showedthat more calcium nodules were formed in HERS-C2 cellsupon the osteogenic induction compared with theHERS-H1 cells (Fig. 3c). Treatment with TGF-β1 signifi-cantly increased the calcium nodule formation especiallyin HERS-H1 cells (Fig. 3c). The quantification analysisconfirmed the significantly higher cementogenic capacityof HERS-C2 cells and TGF-β1 treatment promoted thecementogenic differentiation (Fig. 3d–f ).To confirm the capacity of cementogenesis of the two

cell lines in vivo, HERS-H1 cells and HERS-C2 cells wereseparately transplanted into renal capsule of adult SD ratswith HA/TCP particles. After 8 weeks transplantation, 3out of 4 HERS-C2 grafts generated mineralizedcementum-like tissue, cementoblast-like cells, and fibroustissue (Fig. 4c). However, there was no cementum-liketissue formation in all four grafts in HERS-H1 group andonly fiber-like structure could be observed (Fig. 4b). Incontrol group, no cementum-like or fiber-like tissues wereobserved (Fig. 4a). Immunohistochemical staining showedthat newly formed tissues of HERS-C2 group were posi-tive expression of DMP1 (Fig. 4f), BSP (Fig. 4i), OCN(Fig. 4l), and Periostin (Fig. 4o). While in HERS-H1 group,

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 5 of 13

the newly formed tissues were negative expression ofDMP1 (Fig. 4e), BSP (Fig. 4h), and OCN (Fig. 4k)except for Periostin (Fig. 4n). These results suggestedthat the two cell lines maintained different biologicalfunction. HERS-C2 could differentiate into cemento-blasts and generate cementum-like tissue in vivowhile HERS-H1 could not.

Induction of DPC differentiation into odontoblasts byHERS-H1 and HERS-C2 cellsTo investigate whether the two HERS cell lines have thecapacity to promote the differentiation of the DPCs asprimary HERS cells, the DPCs were treated with theconditioned medium (CM) of the two HERS cell lines

and the odontoblastic differentiation marker genes suchas DMP1 and DSPP were detected. There was a signifi-cant increase in expression level of DMP1 in DPCs aftertreatment with CM of HERS-H1 for 3 days or 7 days, theDSPP expression was also significantly increased aftertreatment with CM of HERS-H1 for 7 days (Fig. 5a).However, DPCs treated with CM of HERS-C2 did notshow a significant increase of DSPP expression either at3 days or at 7 days (Fig. 5a), and the expression of DMP1was increased only after treated with CM of HERS-C2for 3 days but not for 7 days treatment (Fig. 5a). To con-firm the induction of odontoblastic differentiation ofDPCs of the two cell lines, we mixed HERS-H1 orHERS-C2 cells with DPCs and transplanted them into

Fig. 2 HERS-C2 and HERS-H1 underwent EMT induced by TGF-β1. a With the control culture media, HERS-C2 and HERS-H1 showed the cubostone-likemorphology of epithelial cells. After 7 days induction by TGF-β1, HERS-C2 and HERS-H1 became elongated. Scale bars: 50 μm. b, c Expression of EMTmarkers, such as E-cadherin, Vimentin, N-cadherin, Twist1, Snail1, and Zeb1 was examined by real-time RT-PCR in b HERS-C2 cells and c HERS-H1 cellsafter 3 days treatment with TGF-β1. (*P < 0.05; **P < 0.01; ***P < 0.001 vs. control)

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 6 of 13

renal capsule of adult SD rats for 8 weeks. In HERS-H1combined with the DPC group, abundant osteodentin-like tissue was observed (Fig. 5c) and these newly formedtissue positively expressed DMP1 and DSP (Fig. 5f, i).However, in DPC group or HERS-C2 combined with DPCgroup, the newly formed osteodentin-like tissue was obvi-ously thinner than that formed in HERS-H1 combinedwith DPC group (Fig. 5b, c), and the expression of DMP1and DSP was also obviously weaker than that in HERS-H1combined with DPC group (Fig. 5e, g, h, and j).

Gene expression profiles of immortalized HERS cell linesand primary HERS cellsTo elucidate the molecular mechanisms responsible forthe functional difference between the HERS-H1 andHERS-C2 cells, we conducted gene expression profilemicroarray on HERS-H1 cells and HERS-C2 cells, andprimary HERS cells were included as a control. To as-sess the validity of the microarray data, real-time quan-titative PCR was performed to compare geneexpression levels between the cell lines and primary

cells. Five genes (Fgf2, Igf1, Integri a1, Ambn, andAmgn) were selected at random from array dataset tovalidate the microarray data. In all cases, RT-PCR datawas consistent with the results of the microarray, sug-gesting that the dataset obtained from the microarrayanalysis accurately reflects gene expression differencesbetween HERS cell lines and primary HERS cells (Add-itional file 4: Figure S5).We first attempted to identify the gene expression dif-

ference between primary HERS cells and HERS-H1 orHERS-C2 cells. Compared to the primary HERS cells,there were 5202 differential expression genes (2706genes were upregulated and 2496 genes were downregu-lated) in HERS-H1 cells and 4777 differential expressiongenes (2512 genes were upregulated and 2265 geneswere downregulated) in HERS-C2 cells respectively(Fig. 6a, b). Specifically, 1702 were upregulated and 1767were downregulated in both HERS-H1 cells andHERS-C2 cells compared to primary HERS cells asshown in the Venn diagram (Fig. 6b). We had validatedthe expression of a number of identified differential

Fig. 3 Cementogenetic differentiation potential of the two cell lines. a, b The expression of cementoblasts markers was examined by real-timeRT-PCR in the primary, HERS-H1, and HERS-C2 cells after 3 and 7 days culture of osteo-medium and combined treatment with TGF-β1 and osteo-medium. c Representative images of Alizarin Red S staining indicated the potential of cemento-differentiation of HERS-H1 and HERS-C2 cells.d, e, f Quantitative analyses of calcium mineralization in d HERS-C2 cells and f HERS-H1 cells; f comparison between HERS-H1 and HERS-C2cells. (*P < 0.05 vs. control group in a, b, d, e; #P < 0.05 vs. osteo-induction group; *P < 0.05 vs. HERS-H1 in f)

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 7 of 13

Fig. 4 In vivo cementogenesis of HERS-H1 and HERS-C2. a After 8 weeks of transplantation, there were few cells grown in the control transplants.b Only a dense mass of periodontal ligament fiber-like tissues (f) and fibroblast-like cells were found and there was no mineralized structureformed in HERS-H1 transplants. An abundance of blood vessels was formed among the tissues (black arrow). c Thin layer of cementum-like tissue(red arrow) and some cementoblast-like cells (green arrow) were generated on the surface of hydroxyapatite/tricalcium phosphate (HA/TCP)particles in HERS-C2 transplants. Periodontal ligament fiber-like tissues (f) were observed among the mineralized structure. Immunohistochemistryanalysis of the HERS-C2 grafts (f, i, l) showed strong positive staining for DMP1, BSP, and OCN in the cementum-like tissues (red arrow) and thecementoblast-like cells (green arrow), compared to the negative staining with the corresponding antibodies in the control transplants (d, g, j),respectively. In HERS-H1 transplants, e DMP1 was totally negative and (h, k) BSP and OCN were positive in only a few cells (cyan arrow).m Periostinwas negative in the control grafts. n, o The fiber-like tissues both in HERS-H1 and HERS-C2 groups were positive for Periostin. p, q, r Negative controls(NC) did not show staining. (HA: HA/TCP particles. Scale bar: 100 μm)

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 8 of 13

expression genes using RT-PCR to be true positives,confirming that the microarray results were of highquality and reliability (data not shown). Gene Ontology(GO) analysis and pathway mapping revealed that thesedifferential expression genes compared to primary HERScells were closely related to the PI3K-Akt signaling path-way, Hippo signaling pathway, TGF-β signaling pathway,and so on (Additional file 5: Figure S3).

More importantly, compared to the HERS-C2 cells,there were a total of 834 genes upregulated and 1236genes downregulated in HERS-H1 cells (Fig. 6c). Basedon Gene Ontology (GO) analysis, the upregulated geneswere mostly related to the tissue development and ossifi-cation (Additional file 6: Figure S4a), and the downgenes mostly related to the cell differentiation (Add-itional file 6: Figure S4b). Moreover, function/pathway

Fig. 5 Induction of DPC differentiation into odontoblasts and osteodentin generation by HERS-H1 and HERS-C2 cells. a Real-time PCR showedconditioned medium of HERS-H1 or HERS-C2 altered the expression of DMP1 and DSPP in DPCs. b, d After 8 weeks of transplantation, thin layersof dentin-like tissue (yellow imaginary line) and unmineralized fibrous tissue (black arrow) was formed in the DPCs and HERS-C2 + DPC transplants.c A mass of osteodentin (d) was generated in HERS-H1 + DPC group. f, i IHC analysis of HERS-H1 + DPC grafts showed strong positive staining ofDMP1 and DSP in the osteodentin and differentiated DPCs. e, h Only a few cells in the control group showed positive staining for DMP1 and DSP(red arrow). g, j In HERS-C2 + DPC transplants, a small number of cells showed positive staining for DMP1 while DSP staining was completely negative.k, l, m Negative controls (NC) did not show staining. (*P < 0.05 vs. control group; CM: conditioned medium; Scale bars: 100 μm)

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 9 of 13

analysis via Kyoto Encyclopedia of Genes and Genomes(KEGG) of the differential expression genes reveals thatmany upregulated genes were associated with signalingpathways for ECM-receptor interaction, focal adhesion,and PI3K-AKT signaling pathway (Fig. 6d); on the otherhand, many downregulated genes were related to cell ad-hesion molecules (Fig. 6e).

DiscussionDuring the development of the tooth root, HERS cellsoccupy a central position in guiding tooth root forma-tion by differentiation into cementoblasts through EMTand induction of odontoblast differentiation throughEMI. Previous studies mainly investigated the functionsand mechanisms of HERS cells in tooth root develop-ment [6, 24]. Based on the mechanism of tooth root de-velopment, we focus on the application of HERS cells intooth root regeneration. However, HERS cells are a tem-porary structure during the tooth development [25], andlimited availability and lifespan of primary HERS cellsposes a significant barrier to biological investigation and

therapeutic use of tooth tissue engineering. In this study,we established two immortalized non-tumorigenic celllines HERS-H1 and HERS-C2 deriving from Hertwig’sepithelial root sheath for further research and provided anew cell resource for studies on tooth tissue engineering.The cell lines should be able to proliferate easily in

culture while preserving their functional phenotype. Inthis study, results on cellular morphology, epithelial ormesenchymal-associated gene expression, and responseto TGF-β1 treatment have shown that HERS-H1 andHERS-C2 cells retain characteristic of primary HERScells. Based on the gene expression profiles and pathwaymapping, we found that, in immortalized HERS-H1 orHERS-C2 cells, the PI3K-Akt signaling pathway was sig-nificantly upregulated and Hippo and TGF-β signalingpathways were significantly downregulated. Previousstudies showed that PI3K-Akt signaling pathway andHippo signaling pathway were closely related to the sur-vival, growth, and proliferation of several of normal stemcells and cancer cells [26–29]. In tooth epithelial stemcells, PI3K-Akt and Hippo signaling pathways were also

Fig. 6 Comparison of gene expression profiles between HERS-H1 and HERS-C2. a Heat map showing differential gene expression of primary HERScells, HERS-H1 cells and HERS-C2 cells. b Venn diagram showing common and up- or downregulated genes between primary HERS cells andHERS-H1 cells or HERS-C2 cells. c Heat map showing differential gene expression between HERS-H1 cells and HERS-C2 cells. d Signaling pathwaymapping showing the signaling pathway involved in the upregulated genes in HERS-H1 cells compared to HERS-C2 cells. e Signaling pathwaymapping showing the signaling pathway involved in the downregulated genes in HERS-H1 cells compared to HERS-C2 cells

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 10 of 13

critical for the cell proliferation and differentiation [30–32]. Moreover, compared to the primary HERS cells,TGF-β signaling pathway was downregulated both inHERS-H1 and HERS-C2 cells. It was confirmed thatTGF-β signaling, in HERS cells, relied on a Smad-dependent mechanism in regulating Nfic expression viaShh signaling to control root development [6, 33]. Dur-ing the tooth root development, TGF-β signaling trig-gered HERS cells differentiating into cementoblast-likecells via EMT [7, 8, 34]. In HERS-H1 and HERS-C2 celllines established in our study, TGF-β signaling wasdownregulated which might be responsible for the inhib-ition of HERS cell differentiation by inhibited EMT. Whenthe cell lines were treated with exogenous TGF-β1, the ex-pression level of mesenchymal markers and transcrip-tional factors was upregulated, indicating that EMTprogram was reactivated. Moreover, exogenous TGF-β1treatment also promoted the differentiation of HERS celllines in vitro and in vivo. In addition, to evaluate the safetyof the cell lines for tooth root regeneration, we testedwhether the cell lines could transform into tumor cells invivo. After 4 weeks of transplantation, xenograft tumorformation was found in SCC-25 group but not inHERS-C2 and HERS-H1 cells. These results indicate thatHERS-C2 and HERS-H1 cell lines may provide goodmodels to study the function of HERS cells and may benew suitable cell resources of tooth root regeneration.HERS cells were constituted by a heterogeneous cell

population [35]. In this study, we surprisingly found thetwo cell lines had unique features; one of them had theability to differentiation into cementum-like cells, whilethe other had the ability to induce the odontoblastic dif-ferentiation of dental mesenchymal cells. As previousstudies identified, HERS cells were capable of formingmineralized matrix and expressing osteogenic geneswhen cultured under osteo-media [15, 36]. Our resultsproved that osteo-inductive condition and the combinedtreatment of TGF-β1 could trigger the two cell lines dif-ferentiating into cementoblasts. In vivo transplant-ation showed that in vitro expanded HERS-C2 cellsgenerated cementum characterized by a layer of alignedcementum-like structure. Nevertheless, HERS-H1 cellscould only generate fibrous tissue without mineralizedstructure formation, suggesting that only a subset of HERScells carried the capacity of forming cementum-like tissue.If the continuity of HERS is disturbed, the differentiation

of root odontoblasts is compromised [9]. Thus, anothercharacteristic of HERS cell heterogeneity is induction ofodontoblast differentiation. Some studies implied that theHERS could induce the differentiation of the odontoblastsand suggested that HERS functions as a signaling center toguide root formation [6, 37]. Our in vitro studies provedthat the CM of HERS-H1 could promote the differentiationof DPCs. To investigate whether the cell lines could induce

the formation of dentin in vivo, we also transplantedHERS-H1 and DPCs with HA/TCP into renal capsule ofadult SD rats. After 8 weeks of transplantation, we found alarge amount of dentin-like tissue formed and the DSP andDMP1 expression was accelerated by the combination ofHERS-H1, indicating that only HERS-H1 possessed thecapacity of promoting the differentiation of DPCs and thiscell line could be used to regenerate dentin and reconstructtooth root as well as the alveolar bone.In present study, we established two novel cell lines de-

rived from Hertwig’s epithelial root sheath, which repre-sented two key features of HERS cells during the toothroot development. In order to clarify the mechanisms con-tributing to the difference between these two cell lines, wecompared the gene expression profiles and found that aserious of signaling pathways was involved such asECM-receptor interaction signaling. The extracellularmatrix (ECM) is defined by the composite accumulationof structural and functional molecules that are secreted bycells of all tissues and arranged in a tissue-specific 3Dultrastructure [38]. ECM consists of a complex mixture ofstructural and functional macromolecules including gly-cosaminoglycans and fibrous proteins (collagen, elastin,fibronection, and lammin). Specific interactions betweencells and the ECM are mediated by transmembrane mole-cules integrins [39]. The expression of integrin showed acell type-specific pattern and the integrin subunits deter-mined which ECM substrates bind to which cells, and ul-timately influences cellular phenotype and function [40].The ECM regulates numerous biological processes, in-cluding angiogenesis, innervation, and stem cell differenti-ation. In fact, previous study has demonstrated that ECMis essential for tissue architecture and development inbone and tooth [41]. However, in dental-derived stem cellsespecially HERS cells, the function of ECM has not re-ceived sufficient attention. In our data, the genes such asCOL6A1, LAMA3, LAMA2, ITGA10, TGA4, and ITGA7related to ECM-receptor interactions were observed to beupregulated and signaling pathway enrichment analysisshowed that ECM-receptor interactions were significantlyupregulated in HERS-H1 cells compared to HERS-C2cells. ECM is not only a product of cellular activity but isalso a regulator of cellular activity [39]. These higher ex-pression genes involved in ECM-receptor interactionsmight mediate the intercellular interaction and communi-cation between epithelia and mesenchyme. These alsomight be a major reason for the HERS-H1 cells showingmore potential to promote the differentiation of DPCsthan HERS-C2 cells. To thoroughly elucidate the specificmechanism, further researches are still in need. Thetwo cell lines possess two functions of HERS epithe-lial cells respectively: epithelial-mesenchymal transi-tion and epithelial-mesenchymal interaction, which isessential for tooth root formation. As for the tooth

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 11 of 13

root tissue regeneration, combination of the two celllines and dental mesenchyme might imitate the nat-ural process of tooth root development. Comparedwith the single mesenchymal stem cell-based ap-proach of tooth tissue regeneration, our cell lines de-rived from epithelia combined with mesenchymalcell-based method could generate more kinds of toothtissues and provide an inductive microenvironmentthat mimics the natural niche for cell growth anddifferentiation.

ConclusionThis study reports two novel immortalized epithelial celllines which represent two key features of HERS cellsduring the tooth root development for the first time.Both two cell lines did not induce tumorigenesis in vivoand proved to be biologically safe for implantation.HERS-C2 could differentiate into cementoblasts whichmay produce cementum. HERS-H1 could promote thedifferentiation and dentin formation of DPCs. The invitro and in vivo experiments also confirm that HERScells are of heterogeneity and suggest that the two celllines provide substitutes for primary cells in biologicallyrelevant applications in the fields of cell biology, devel-opmental biology, and tooth tissue engineering, whichconstitute a novel and promising approach for futureclinical dental therapy.

Additional files

Additional file 1: Table S1. Primer sequences used for quantitative real-time PCR. (DOCX 15 kb)

Additional file 2: Figure S1. The examination of tumorigenicity of twoimmortalized cell lines. (A) Macroscopic appearance and (B) tissue sectionsof HE stain showed injection of primary HERS cells, HERS-C2 cells, HERS-H1cells into nude mice did not formation tumor after 4 weeks; tumorformation was observed in positive control group SCC-25 cells (redimaginary line). Scale bars: 200 μm. (TIF 12083 kb)

Additional file 3: Figure S2. Seven days induction by TGF-β1 of HERS-H1 cells. The expression level of E-cadherin was downregulated and thelevel of vimentin, N-cadherin and snail1 upregulated examined by real-timeRT-PCR. (*P < 0.05; **P < 0.01 vs. control). (TIF 587 kb)

Additional file 4: Figure S5. Confirmation of microarray data usingreal-time RT-PCR. Real-time RT-PCR was carried out to validate the arrayresults. Total RNA from three independent cultured cells were used forthe analysis. Triplicate assays were performed from each RNA sample.Data are normalized using GAPDH as an endogenous control for RNA input.(TIF 5232 kb)

Additional file 5: Figure S3. Signaling pathway mapping of HERS-H1cells and HERS-C2 cells. (A) Signaling pathway of upregulated genes ofHERS-H1 vs. primary HERS cells. (B) Signaling pathway of downregulatedgenes of HERS-H1 vs. primary HERS cells. (C) Signaling pathway of upregulatedgenes of HERS-C2 vs. primary HERS cells. (D) Signaling pathway ofdownregulated genes of HERS-C2 vs. primary HERS cells. (TIF 1013 kb)

Additional file 6: Figure S4. Go analysis of the differential expressiongenes. (A) Function analysis of the upregulated genes of HERS-H1 cellscompared to HERS-C2 cells was conducted via GO analysis. (B) Functionanalysis of the downregulated genes of HERS-H1 cells compared toHERS-C2 cells was conducted via GO analysis. (TIF 602 kb)

AbbreviationsCM: Conditioned medium; DPCs: Dental papilla cells; EMI: Epithelial–mesenchymal interaction; EMT: Epithelial–mesenchymal transition;HA: Hydroxyapatite/tricalcium phosphate; HERS: Hertwig’s epithelial rootsheath

AcknowledgementsNot applicable.

FundingThis work was supported by National Key Research and DevelopmentProgram of China (NO.2017YFA0104800).

Availability of data and materialsAll data generated or analyzed during this study are included in thispublished article and its supplementary information files.

Authors’ contributionsXL contributed to the conception and design, collection and/or assembly ofdata, data analysis and interpretation, and manuscript writing; SZ contributedto the conception and design and collection and/or assembly of data; ZZand WG contributed to the collection and/or assembly of data; GCcontributed to the conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, andfinal approval of the manuscript; WT contributed to the conception anddesign, financial support, and final approval of the manuscript. All authorshave reviewed the manuscript before submission. All authors read andapproved the final manuscript.

Ethics approval and consent to participateAll experiments were conducted in accordance with a protocol approved bythe Committee of Ethics of West China Hospital of Stomatology of SichuanUniversity (NO. WCHSIRB-D-2018-100).

Consent for publicationAll authors have contributed to, read, and approved the manuscript forsubmission.

Competing interestsThe authors declare that they have no competing interests.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Author details1State Key Laboratory of Oral Disease, West China Hospital of Stomatology,Sichuan University, Chengdu, China. 2National Engineering Laboratory forOral Regenerative Medicine, West China Hospital of Stomatology, SichuanUniversity, Chengdu, China. 3National Clinical Research Center for OralDiseases, West China Hospital of Stomatology, Sichuan University, Chengdu,China. 4Department of Oral and Maxillofacial Surgery, West China Hospital ofStomatology, Sichuan University, Chengdu, China. 5Department of Pediatric,West China Hospital of Stomatology, Sichuan University, Chengdu, China.6West China School of Public Health, Sichuan University, Chengdu, China.

Received: 11 August 2018 Revised: 9 November 2018Accepted: 10 December 2018

References1. Volponi AA, Pang Y, Sharpe PT. Stem cell-based biological tooth repair and

regeneration. Trends Cell Biol. 2010;20:715–22.2. Tatullo M, Marrelli M, Shakesheff KM, et al. Dental pulp stem cells: function,

isolation and applications in regenerative medicine. J Tissue Eng RegenMed. 2015;9:1205–16.

3. Guo W, Gong K, Shi H, et al. Dental follicle cells and treated dentin matrixscaffold for tissue engineering the tooth root. Biomaterials. 2012;33:1291–302.

4. Bluteau G, Luder HU, De Bari C, et al. Stem cells for tooth engineering. EurCells Mater. 2008;16:1–9.

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 12 of 13

5. Wentz FM, Weinmann JP, Schour I. The prevalence, distribution, andmorphologic changes of the epithelial remnants in the molar region of therat. J Dent Res. 1950;29:637–46.

6. Huang X, Xu X, Bringas P, et al. Smad4-Shh-Nfic signaling Cascade-mediatedepithelial-mesenchymal interaction is crucial in regulating tooth rootdevelopment. J Bone Miner Res. 2009;25:1167–78.

7. Sonoyama W, Seo B-M, Yamaza T, et al. Human Hertwig’s epithelial root sheathcells play crucial roles in cementum formation. J Dent Res. 2007;86:594–9.

8. Chen J, Chen G, Yan Z, et al. TGF-β1 and FGF2 stimulate the epithelial-mesenchymal transition of HERS cells through a MEK-dependentmechanism. J Cell Physiol. 2014;229:1647–59.

9. Kim TH, Bae CH, Lee JC, et al. β-Catenin is required in odontoblasts fortooth root formation. J Dent Res. 2013;92:215–21.

10. Iohara K, Nakashima M, Ito M, et al. Dentin regeneration by dental pulpstem cell therapy with recombinant human bone morphogenetic protein 2.J Dent. 2004:590–5.

11. Yang X, Yang F, Walboomers XF, et al. The performance of dental pulp stemcells on nanofibrous PCL/gelatin/nHA scaffolds. J Biomed Mater Res - Part A.2010;93:247–57.

12. Oka S, Oka K, Xu X, et al. Cell autonomous requirement for TGF-β1 signalingduring odontoblast differentiation and dentin matrix formation. Mech Dev.2007;124:409–15.

13. Suchetha A, Spandana A, Sapna N, et al. Hertwig ’ s epithelial root sheath :a panaromic view. Int J Appl Sci. 2017;3:21–5.

14. Nam H, Kim J-H, Kim J-W, et al. Establishment of Hertwig’s epithelial rootsheath/epithelial rests of Malassez cell line from human periodontium. MolCells. 2014;37:562–7.

15. Zeichner-David M, Oishi K, Su Z, et al. Role of Hertwig’s epithelial rootsheath cells in tooth root development. Dev Dyn. 2003;228:651–63.

16. Chen G, Sun W, Liang Y, et al. Maternal diabetes modulates offspring cellproliferation and apoptosis during odontogenesis via the TLR4/NF-κBsignalling pathway. Cell Prolif. 2017;50:1–12.

17. Yang Y, Ge Y, Chen G, et al. Hertwig’s epithelial root sheath cells regulateosteogenic differentiation of dental follicle cells through the Wnt pathway.Bone. 2014;63:158–65.

18. Chen G, Chen J, Yan Z, et al. Maternal diabetes modulates dental epithelialstem cells proliferation and self-renewal in offspring through apurinic/apyrimidinicendonuclease 1-mediated DNA methylation. Sci Rep. 2017;7:1–12.

19. Xing Y, Nie X, Chen G, et al. Comparison of P75NTR-positive and -negativeetcomesenchymal stem cell odontogenic differentiation through epithelial-mesenchymal interaction. Cell Prolif. 2016;49:185–94.

20. Berendsen AD, Smit TH, Schoenmaker T, et al. Inorganic phosphatestimulates DMP1 expression in human periodontal ligament fibroblastsembedded in three-dimensional collagen gels. Cells Tissues Organs. 2010;192:116–24.

21. Cao Z, Zhang H, Zhou X, et al. Genetic evidence for the vital function ofosterix in cementogenesis. J Bone Miner Res. 2012;27:1080–92.

22. Foster BL. Methods for studying tooth root cementum by light microscopy.Int J Oral Sci. 2012;4:119–28.

23. Salmon CR, Giorgetti APO, Paes Leme AF, et al. Global proteome profiling ofdental cementum under experimentally-induced apposition. J Proteome.2016;141:12–23.

24. Xu H, Snider TN, Wimer HF, et al. Multiple essential MT1-MMP functions intooth root formation, dentinogenesis, and tooth eruption. Matrix Biol. 2016;52–54:266–83.

25. Shinmura Y, Tsuchiya S, Hata KI, et al. Quiescent epithelial cell rests of malassezcan differentiate into ameloblast-like cells. J Cell Physiol. 2008;217:728–38.

26. Dreesen O, Brivanlou AH. Signaling pathways in cancer and embryonicstem cells. Stem Cell Rev. 2007;3:7–17.

27. Hansen CG, Moroishi T, Guan KL. YAP and TAZ: a nexus for hippo signalingand beyond. Trends Cell Biol. 2015;25:499–513.

28. Zanconato F, Cordenonsi M, Piccolo S. YAP/TAZ at the roots of Cancer.Cancer Cell. 2016;29:783–803.

29. Zhu J, Qin J, Shen Z, et al. Dispase rapidly and effectively purifies Schwanncells from newborn mice and adult rats. Neural Regen Res. 2012;7:256–60.

30. Arai C, Yoshizaki K, Miyazaki K, et al. Nephronectin plays critical roles in Sox2expression and proliferation in dental epithelial stem cells via EGF-likerepeat domains. Sci Rep. 2017;7:1–12.

31. Kwon HJE, Li L, Jung HS. Hippo pathway/yap regulates primary enamel knotand dental cusp patterning in tooth morphogenesis. Cell Tissue Res. 2015;362:447–51.

32. Wang J, Martin JF. Hippo pathway: an emerging regulator of craniofacialand dental development. J Dent Res. 2017;96:1229–37.

33. Li J, Parada C, Chai Y. Cellular and molecular mechanisms of tooth rootdevelopment. Development. 2017;144:374–84.

34. Xiong J, Mrozik K, Gronthos S, et al. Epithelial cell rests of Malassez containunique stem cell populations capable of undergoing epithelial-mesenchymal transition. Stem Cells Dev. 2012;21:2012–25.

35. Huang X, Bringas P, Slavkin HC, et al. Fate of HERS during tooth rootdevelopment. Dev Biol. 2009;334:22–30.

36. Ten Cate a R. The role of epithelium in the development, structure andfunction of the tissues of tooth support. Oral Dis. 1996;2:55–62.

37. Thomas HF, Kollar EJ. Differentiation of odontoblasts in graftedrecombinants of murine epithelial root sheath and dental mesenchyme.Arch Oral Biol. 1989;34:27–35.

38. Mouw JK, Ou G, Weaver VM. Extracellular matrix assembly: a multiscaledeconstruction. Nat Rev Mol Cell Biol. 2014;15:771–85.

39. Hussey GS, Keane TJ, Badylak SF. The extracellular matrix of thegastrointestinal tract: a regenerative medicine platform. Nat RevGastroenterol Hepatol. 2017;14:540–52.

40. Kim SH, Turnbull J, Guimond S. Extracellular matrix and cell signalling: thedynamic cooperation of integrin, proteoglycan and growth factor receptor.J Endocrinol. 2011;209:139–51.

41. Ravindran S, George A. Multifunctional ECM proteins in bone and teeth. ExpCell Res. 2014;325:148–54.

Li et al. Stem Cell Research & Therapy (2019) 10:3 Page 13 of 13